European Polymer Journal 173 (2022) 111256

Contents lists available at ScienceDirect

European Polymer Journal

journal homepage: www.elsevier.com/locate/europolj

3D printable hybrid acrylate-epoxy dynamic networks

J. Casado a, O. Konuray a, *, A. Roig b, X. Fernández-Francos a, X. Ramis a
a
Thermodynamics Lab, ETSEIB – UPC, Avda. Diagonal 647, 08028 Barcelona, Spain
b
Dept. of Analytical and Organic Chemistry, Universitat Rovira I Virgili, C/ Marcel⋅lí Domingo, 1. 43007 Tarragona, Spain

A R T I C L E I N F O A B S T R A C T

Keywords: Polymer networks with dynamic bonds, also known as covalent adaptable networks (CANs) combine the superior
Covalent adaptable network mechanical properties and chemical resistance of thermosets with the ability to be reprocessed, a feat formerly
Vitrimer attributed only to thermoplastics. Inspired by an evergrowing body of research on dynamic poly(β-hydroxy ester)
3D printing
networks, a polyacrylate/epoxy-acid thermoset was designed, which contains a high concentration of β-hydroxy
Beta-hydroxy ester
Dual-cure
ester bonds that partake in transesterification reactions that facilitate the repair and recycle of the cured material
at feasible temperatures. Firstly, liquid formulations are subjected to UV light to initiate acrylate homo­
polymerization to obtain the intermediate, partially-cured material. A subsequent thermal treatment triggers the
epoxy-acid reaction, which improves the mechanical properties and helps increase the likelihood of trans­
esterifications as new β-hydroxy ester groups are formed. The effect of thermal post-treatment and the choice of
catalyst on viscoelastic properties and stress relaxation behavior of these materials is studied. Results show that,
transesterification reactions reach equilibrium in less than 4 h at 180 ◦ C during which time the overall cross-
linking density increases further. As to the choice of catalyst, a commonly used zinc acetylacetonate out­
performs an imidazole-type base. Thanks to the dynamic bonds, damaged samples can be repaired fully using
simple procedures. Recyclability is tested by grinding pristine samples and re-molding them under pressure and
temperature. Practically complete recovery of viscoelastic properties is confirmed.

1. Introduction underwent transesterification when heated [7].

Covalent adaptable networks (CANs) are thermosetting polymers
featuring covalent bonds that are able to reshuffle through exchange
reactions triggered using appropriate stimuli, such as heat. Given that
the overall network connectivity is maintained during these exchange
reactions, the thermoset acquires a certain malleability and reparability
when heated, which are abilities formerly attributed exclusively to
thermoplastics. The definition of CANs was introduced more than a
decade ago by Kloxin et al. [1] and possible exchange mechanisms that
underlie CANs have been reviewed in detail [2]. Furthermore, there are
excellent reviews that treat the subject of CANs at the material level,
with consideration of mechanical properties [3,4] and also with respect
to possible applications such as biomaterials and additive
manufacturing [5,6]. As a matter of fact, CANs that can maintain their
network connectivity while undergoing exchange reactions, and thus
exhibit an Arrhenius-like flow behavior when heated, were given the
name “vitrimers”, since this behavior is akin to vitreous silica. The first
reported vitrimer was a poly(hydroxy ester) based network which
In the realm of ester based dynamic networks, zinc salts are
commonly employed as transesterification catalysts. An epoxy-acid
reprocessable material with zinc acetoacetate as catalyst was
described by Yu et al wherein they study the effect of reprocessing
conditions (such as temperature, time, and pressure), and multiple
reprocessing on mechanical properties [8]. In another work, Niu et al
copolymerized vinyl monomers to obtain carboxyl and zinc carboxylate
containing pre-polymers that were later post-cured using DGEBA [9].
The Zn2+ ions catalyze transesterification, enabling reprocessability.
The catalysis of transesterification by zinc salts was also documented in
epoxy vitrimers [7,10,11] or other β-hydroxy ester containing polymers
[12]. The presence of hydroxyl groups in the matrix is reported to
enhance transesterification rate. In fact, neighboring β-hydroxyls
enhance reprocessability up to 10 times [13], apparently through a
mechanism in which the hydrogen bond forming in the β-hydroxyl
containing structures renders the carbonyl carbon more electrophilic,
thereby facilitating attack by the electrophile such as alcohol (R-OH)
[14]. In the extreme case (i.e. when they are abundant) they might

* Corresponding author.
E-mail address: ali.osman.konuray@upc.edu (O. Konuray).

https://doi.org/10.1016/j.eurpolymj.2022.111256
Received 11 March 2022; Received in revised form 28 April 2022; Accepted 5 May 2022
Available online 10 May 2022
0014-3057/© 2022 The Authors. Published by Elsevier Ltd. This is an open access article under the CC BY license (http://creativecommons.org/licenses/by/4.0/).
J. Casado et al.
alleviate altogether the need for a catalyst. One such example of a
catalyst-free epoxy vitrimer was demonstrated by Han et al [15].
CANs synthesized from photocurable monomers have been evalu­
ated in stereolithography based 3D printing applications which
conventionally rely on non-recylable thermosets. The introduction of
CANs thus can help alleviate the carbon footprint of these 3D printed
products. While most stereolithographically 3D printed CANs reported
so far are based on acrylate photopolymerization [12,16], some are
based on thiol-acrylate click chemistry [17,18]. Two-component CANs
printed by purely thermal, extrusion-based methods are also reported,
along with many dual-cure and 3D printable CANs [19]. In this latter
group of materials, an acrylate based component is firstly cured in a
stereolithography-based process, such as in a digital light processing
(DLP) printer. After that the printed object is thermally treated to cure a
secondary monomer in the system which can be, among other possi­
bilities, an epoxide [20] or a polyurethane [21].
Inspired by all the above, in this work, we have formulated a hybrid
acrylate-epoxy CAN that can be printed in a DLP printer. The process
starts with a diacrylate monomer, namely glycerol 1,3-diglycerolate
diacrylate (GLYDA), which undergoes radical-mediated acrylate
homopolymerization to yield a tightly crosslinked polyacrylate network.
The GLYDA molecule features two β-hydroxy ester groups and addi­
tional pendant hydroxyl groups along its aliphatic backbone. To the best
of our knowledge, the use of a β-hydroxy ester containing acrylate in a
hybrid CAN was not reported before. When the fully cured material is
thermally treated, various transesterification reactions are expected to
occur. These ester-ester or ester-hydroxyl exchanges can take place
across all the β-hydroxy ester moieties present in the overall polymeric
network. These moieties either originate from GLYDA itself, or form as a
result of epoxy-acid reaction. The DLP cured objects are subsequently
subjected to elevated temperatures to cure the epoxy component.
Similar to other dual-cure systems previously described by our group,
the properties at each curing stage can be tailored by changing monomer
composition, according to processing and application requirements
[22,23]. Following a well-established strategy to prepare vitrimers, two
catalysts capable of promoting both the epoxy-acid reaction and the
transesterification were used: a zinc acetylacetonate, Zn(AcAc)2 [24],
and 1-methylimidazole [25]. After a full characterization of curing ki­
netics and viscoelastic properties of the cured materials, we investigate
Scheme 1. Monomers catalysts and initiators used in the formulations.
2
European Polymer Journal 173 (2022) 111256
the effect of catalyst and of thermal post-cure on the final viscolelastic
properties and stress-relaxation behavior.
2. Materials and methods
The molecular structures of the chemicals used are given in Scheme
1. The formulations studied are given in Table 1.
All chemicals, except the epoxy resin (ER), were purchased from
Sigma-Aldrich (Madrid, Spain) and used without purification. ER was
kindly supplied by Po.Int.Er S.R.L. (Valfenera, Italy). Mixtures con­
taining Zn(AcAc)2 were prepared by dissolving the transesterification
catalyst and photoinitiator (either TPO or DMPA) in the acrylate
monomers at 60 ◦ C, and then by adding ER and GLU at the same tem­
perature. Mixtures were magnetically agitated until homogenization.
When 1-MI was used, it was added at the end of the preparation after
cooling down the mixture. Slightly turbid solutions were obtained in
formulations with Zn(AcAc)2 due possibly to its inherent incompatibility
with the diacrylate. GMA was used as coupling agent between the pol­
yacrylate and epoxy-acid networks to improve homogeneity. GMA/
GLYDA acrylate molar ratio was 1:1. The molar ratio between acid
groups (GLU) and epoxide groups (GMA + ER) was also 1:1. A different
mixture, named Base-3D, was prepared by modifying the basic formu­
lation with PEGMA (MW = 300 g mol− 1), such that the PEGMA:GLYDA
weight ratio was 1:1. This was done to reduce viscosity so that the resin
can be easily 3D printed. The addition of PEGMA was observed to pro­
mote the dispersion of Zn(AcAc)2 during the preparation of the formu­
lations and had a positive effect on network homogeneity as well (vide
infra). The first curing stage was carried out using either of the following
two methods: 1) using a Vilber-Lourmat UV oven and hand-made
transparent molds with PTFE spacers, or 2) using a Asiga MAX UV385
DLP-3D printer (Base-3D). The second curing stage was carried out at
120 ◦ C for 12 h in a Memmert convection oven to ensure complete cure.
Thermal post-treatments at 180 ◦ C were carried out also in the same
oven.
A Mettler DSC3 + calorimeter was used to measure both the Tg and
the polymerization heats. Photocuring experiments were performed
using a Hamamatsu LC5 light source equipped with a Hg-Xe mid-pres­
sure lamp adapted to a Mettler DSC821 calorimeter. UV light intensity
was 36 mW cm− 2 measured at 365 nm using a radiometer. Conversion of
a
MW = 300 g mol− 1, b
Epoxy equivalent wt. = 187 g ee-1.
J. Casado et al. European Polymer Journal 173 (2022) 111256

Table 1
Composition of formulations based on 1 g of GLYDA. Catalyst loadings are given as wt. percentages.
Formulation GLYDA (g) PEGMA (g) GMA (g) ER (g) GLU (g) PI* (2%) (g) Zn(AcAc)2 (6%) (g) 1-MI (1%) (g)

Base-Zn 1.00 – 0.82 2.00 1.09 0.10 0.29 –

Base-MI 1.00 – 0.82 2.00 1.09 0.10 – 0.05
Base-Mix 1.00 – 0.82 2.00 1.09 0.10 0.29 0.05
Base-3D 1.00 1.00 0.82 2.00 1.09 0.12 0.35 –

*PI: Photoinitiator; either TPO or DMPA.

functional groups, defined as a function of time x(t), was calculated
using x(t) = Δh(t)/Δhtotal where Δh(t) is the heat evolved up to time t,
and Δhtotal is the total polymerization heat. To probe the storage stability
of intermediate (i.e. only photocured) samples, a different equation,
namely x = 1 − Δhresidual /Δhtotal was used where Δh is the heat evolved
during the cure of a sample stored for a known duration of time, and
Δhtotal is the total reaction heat of a freshly prepared and photocured
sample.
The intermediate and final (i.e. fully cured material) Tg were taken as
the half-way point of the heat capacity step observed in a scan at 10 ◦ C
min− 1, following the DIN 51,007 standard method.
Functional group conversions were also monitored using a
temperature-controlled Brucker Vertex 70 FTIR spectrometer equipped
with an attenuated total reflection (ATR) accessory (GoldenGate™,
Specac Ltd.). The relevant wavelengths are marked on the spectra pre­
sented (vide infra). Spectra were collected in absorbance mode with a
resolution of 4 cm− 1 within the 600–4000 cm− 1 wavelength range,
averaging 20 scans for each spectrum. In photocuring tests, the same UV
irradiation equipment with the same specifications as in DSC measure­
ments was used.
A TA Instruments DMA Q800 device was used for thermomechanical
analysis of cured samples. Prismatic samples with dimensions 1.5 × 10
× 30 mm3 (thickness × width × length) were analyzed using a single
cantilever clamp with a free length of 10 mm at a frequency of 1 Hz and
amplitude of 15 µm at 3 ◦ C min− 1 from − 50 ◦ C up to the rubbery state to
measure storage modulus, loss modulus, and tan delta. Stress relaxation
experiments were performed on samples with dimensions 1.5x10x20
mm3 using a 3-point bending configuration with a preload force of 0.01
N and a fixed strain of 1%. Flexural modulus was determined at 25 ◦ C
using a 3-point bending configuration with the same dimensions as in
stress relaxation experiments, using the same preload force and a force
ramp of 3 N/min.
Thermogravimetric analysis (TGA) was performed at 10 ◦ C/min
using a Mettler TGA/SDTA 851e/LF/1100 thermobalance. Fully cured
samples were analyzed under isothermal conditions and under 50 cm3
min− 1 of nitrogen purge.
Reprocessing capability of the fully cured material was tested by
finely chopping a sample and hot-pressing in a Specac Atlas manual 15 T
hydraulic press at 9.25 MPa in an aluminum mold at about 180 ◦ C for 5
h.
3. Results and discussion
3.1. Characterization of curing kinetics
FTIR spectra collected during the photocuring of Base-3D is given in
Fig. 1. The disappearance of the acrylate and methacrylate bands at
various wavelengths (marked on the figure) and the intactness of the
epoxy band at 915 cm− 1 confirms complete acrylate conversion (stage 1
curing) and no epoxy-acid reaction. Moreover, the carbonyl band at
1720 cm− 1 (acrylate C=O) shifts to 1730 cm− 1 (ester C=O), making the
acid carbonyl characteristic peak at 1720 cm− 1 visible, similar to what
was observed in earlier works [26]. These observations suggest the only
reaction taking place is the radical-mediated acrylate/methacrylate
polymerization.
In Fig. 2, the FTIR spectra collected during thermal postcuring of a

Fig. 1. FTIR spectra of the photocuring of Base-3D with UV light at 30 ◦ C. Full acrylate + methacrylate conversion is achieved in 30 s.

3
J. Casado et al. European Polymer Journal 173 (2022) 111256

Fig. 2. Thermal (stage 2) curing of Base-3D at 120 ◦ C monitored in FTIR. Hydoxyl formation due to new β-hydroxy ester bonds, and acid hydroxyl consumption is
observed above 3000 cm− 1 band (left) along with the disappearance of epoxy groups (right). 96% conversion is achieved in 180 min.

UV-cured Base-3D formulation at 120 ◦ C can be seen. Practically com­
plete conversion is achieved in 3 h. The disappearance of the carboxylic
acid OH band and the appearance of the β-hydroxy ester OH band,
together with the disappearance of the epoxy group, indicate that the
second curing proceeds via the expected epoxy-acid reaction.
The first curing stage is complete in mere seconds (Fig. 1), while the
second curing stage occurs at higher temperatures and at a rate several
orders of magnitude lower. Globally, the curing takes place in two well-
defined, sequential and orthogonal stages, similar to other 3D-printable
dual-curing systems [27,28].
Comparing the curing rates of the two stages, one can confirm that
thermal curing is the rate determining step in the processing of these
materials. Although a fast curing reaction might prove practical, it might
also jeopardize storage stability. The epoxy-acid reaction is not latent
and thus a slow second curing stage might actually be crucial in terms of
storage stability of the uncured mixtures and also intermediate stage
materials. It is therefore interesting to study the effect of catalyst choice
and formulation composition on this second curing stage.
In Fig. 3, DSC thermograms at various isotherms for the second
curing stage of Base-3D are given. FTIR and DSC conversions are
compared at 120 ◦ C and a good fit is observed as can be seen. This
suggests that the degree of cure can be determined accurately with
either technique. In spite of the relatively slow curing kinetics, the
second curing completes in approx. 1 hr at 140 ◦ C.
To elucidate the kinetics of the epoxy-acid reaction, DSC data were
analyzed using i) an integral isoconversional method and, ii) model-
fitting. For the latter, an autocatalytic, Kamal-type model with a total
reaction order of 2 was used (i.e. n + m = 2), similar to other epoxy-
acid/anhydride reactions modeled earlier [29]. The reader is directed
to the same reference for detailed explanations of both kinetic analysis
methods. The isothermal isoconversional method is based on the
assumption that reaction rate at a given conversion is solely a function of
temperature, leading to the following linear expression.

Fig. 3. Isothermal curing profiles corresponding to the thermal (stage 2) curing of Base-3D at different temperatures. The samples were previously photocured
at 30 ◦ C.

4
J. Casado et al. European Polymer Journal 173 (2022) 111256

( )
g(α) Eα containing formulations (Base-Mix and Base-MI) are higher than those
lnti,α = ln + (1)
Aα R⋅Ti of Zn(AcAc)2 containing formulations (Base-Zn and Base-3D) by an
order of magnitude. Regarding reaction orders, the values are compa­
where Aα is the pre-exponential factor, Eα is the activation energy and ti,α rable across different formulations, suggesting a single, catalyst-
the time required to achieve a fixed conversion at temperature Ti and independent curing mechanism. Nevertheless, as will be shown later,
g(α) is the integral model function defined as. the choice of catalyst brings about a trade-off between the rate of curing
and the rate of relaxation (vide infra).
∫α
dα The results of the kinetics analysis suggest also that the storage sta­
g(α) = (2)
f (α) bility of the prepared formulations and the intermediate stage material
would be much higher in the case of the formulations containing only Zn
0

where f(α) is the differential model function. If the separable function (AcAc)2. The time t needed to reach a 5 % conversion at 30 ◦ C was
extrapolated making use of the kinetic parameters in Table 2 and
assumption is correct, a plot of lnti,α versus (R⋅Ti )− 1 i, gives a straight
expression g(α) = k⋅t, where k = A⋅exp( − E/RT) is the rate constant. For
with slope Eα and y-intercept ln(g(α)/Aα ). If the the model function (g(α)
the Base-3D formulation, the estimated time was 42 h, and for Base-Mix
or f(α) is known, the pre-exponential factor can easily be calculated for
it was less than 6 h. Therefore, it can be projected that using Zn(AcAc)2
each conversion. To validate the model, equation (1) is rearranged to
would guarantee sufficient storage stability. For experimental verifica­
give.
tion of this, intermediate stage materials were kept in storage at 30 ◦ C
( )
ti,α E for prolonged periods to check their stability. The samples were peri­
ln = − ln(A) + (3)
g(α) R⋅Ti odically tested in DSC to monitor the evolution of Tg and residual heats.
( ) As can be seen in Fig. 5, the presence of 1-MI jeopardizes the storage
Plotting ln ti,α /g(α) against (R⋅Ti )− 1 for all conversions will yield a stability, as the Tg increases from − 27 ◦ C to − 12 ◦ C in only 2 days (Tg
straight line with a slope E, and y-intercept –ln(A) if the kinetic model is reading on day 2 is marked with blue dashed line in Fig. 5), while the
correct. residual heat drops by 46%. The Tg reaches 28 ◦ C in about one month.
The Kamal model function and its integral are given in equations (4) On the other hand, when Zn(AcAc)2 is the catalyst, storage stability is
and (5), respectively. remarkable. The increase in Tg is below 10 ◦ C even after 10 days. In spite
f (α) = αm (1 − α)n (4) of the inherent extrapolation error, the kinetic model predictions are in
reasonable agreement with experimental findings.
1 ( α )1− m In spite of the good storage stability of its formulations, the solubility
g(α) = (5)
1− m 1− α of Zn(AcAc)2 can be a matter of concern. Compatibilization of zinc salts
As mentioned previously, it is assumed that n + m = 2, where n and might be complicated unless there is sufficient concentration of hy­
m are the partial reaction orders. In our analysis, the parameters n and m droxyls and/or carboxyls in the medium to facilitate desirable ionic
were calculated using the GRG Nonlinear algorithm of MS Excel. The interactions. As stated previously, even though the formulations herein
algorithm attempts to maximize the R-squared value of the linear prepared have high hydroxyl and carboxyl contents, slightly turbid so­
regression which also gives the activation energy E and the logarithm of lutions were obtained when Zn(AcAc)2 was used. In contrast, the 1-MI
the pre-exponential factor A (vide supra). The resulting kinetic param­ catalyst, being in liquid form, was much easily dissolved during prepa­
eters are tabulated in Table 2. The values of E were practically identical ration of formulations. Nevertheless, the lower storage stability of its
to the apparent activation energy obtained using the isoconversional formulations is a caveat.
integral method. It can also be observed that the activation energies and
reaction orders are very similar for all formulations. This indicates that
3.2. Thermomechanical properties
the epoxy-acid reaction proceeds via the same mechanism regardless of
other formulation components and the employed catalyst. The most
Table 3 summarizes glass transition temperatures Tg of the materials
noticeable difference is the value of the pre-exponential factor and the
at the two curing stages, as well as after further thermal treatment at
calculated kinetic constant at 120 ◦ C. The fastest curing system is Base-
180 ◦ C for 4 h, which was done to ensure that transesterification re­
MI. When Zn(AcAc)2 is used as catalyst, the curing reaction slows down.
actions attained equilibrium. Similar thermal treatment procedures
Accordingly, both Base-Zn and Base-3D have slow and comparable
were employed by other researchers as well [12,31].
curing kinetics. This difference in reactivity is also illustrated in Fig. 4. It
The choice of catalyst did not seem to alter the Tg at any curing stage.
can be seen that the effect of PEGMA in the Base 3D formulation is
However, the modification by PEGMA seems to have slightly decreased
practically insignificant since complete conversion is attained in about 3
Tg at all stages, possibly due to a decrease in crosslinking density along
h regardless of the presence of PEGMA.
with an increase in free volume caused by the presence of the aliphatic
The interpretation of the kinetic parameters is not straightforward
side chain of PEGMA. The thermal post-treatment seems to have a
due to the fact that in autocatalytic curing reactions, E and A tend to
beneficial effect in terms of Tg , especially in PEGMA-free formulations.
compensate each other. As a result, referring to either parameter alone
As will be shown later, at 180 ◦ C, transesterification occurs at rapid
would be misleading in drawing conclusions about reaction rate. How­
rates, giving way to new structural fragments containing beta-hydroxy
ever, the rate constant k, being a combination of the two, represents
esters, diols, and diesters. As will be shown, the overall network con­
curing kinetics in a more reliable way [30]. Corroborating what was
nectivity appears to increase as a result of these new structures. It was
observed experimentally, the rate constant at 120 ◦ C (k120◦ c) of 1-MI
seen that treatments longer than 4 h did not increase Tg any further (vide
infra). Fully cured samples (without post-treatment) were analyzed in
Table 2 DMA and results are given in Fig. 6. Although there was no visual dif­
Kinetic parameters obtained by model-fitting and autocatalytic model function.
ference between Base-3D and the rest of the (PEGMA-free) formulations,
k120◦ c was calculated using Arrhenius eqn.
the PEGMA-free samples exhibited bimodal tan delta curves regardless
Formulation n m E(kJ/mol) lnA(min(− 1)
) k120◦ C (min− 1 ) of the catalyst used. In spite of the acrylate/methacrylate and the epoxy-
Base-Zn 1.75 0.25 71 19.02 0,0671 acid interconnection (facilitated by the coupling agent GMA), the
Base-Mix 1.76 0.24 72 20.75 0,2787 overall network appears to have a heterogeneous structure, possibly due
Base-MI 1.68 0.32 73 21.41 0,3972 to the tightly crosslinked polyacrylate network impeding the homoge­
Base-3D 1.65 0.35 72 19.26 0,0628
neous penetration of the epoxy-acid network into it. The narrow tan

5
J. Casado et al. European Polymer Journal 173 (2022) 111256

Fig. 4. Isothermal DSC heat flow and conversion curves of all formulations at 120 ◦ C.

Fig. 5. DSC measured residual curing heats (hollow symbols) and Tg of intermediate materials (filled symbols) showing the effect of catalyst on storage stability. A
temperature ramp of 10 ◦ C min-1 was used in heat scans to determine residual heats. Tg reading on day 2 is marked with blue dashed arrow. (For interpretation of the
references to colour in this figure legend, the reader is referred to the web version of this article.)

should be due to the acrylate homopolymer phase (pure GLYDA polymer

Table 3
has a calorimetric Tg of 85 ◦ C, which would correspond to a tan delta
Tg of formulations at different stages of cure/processing. Materials attain stage 1
peak around 100–110 ◦ C). Similar effects were reported in other dual-
(S1) after being photocured at 30 ◦ C, stage 2 (S2) after thermal cure at 120 ◦ C for
12 h, and final stage after thermal post-treatment at 180C for 4 h. cured acrylate/epoxy-anhydride 3D-printable formulations [27,28],
The Base-3D material exhibited a narrower relaxation, possibly due to a
Formulation Tg,S1 (◦ C) Tg,S2 (◦ C) Tg,final (◦ C)
more homogeneous, less tightly cross-linked acrylate network facilitated
Base-Zn − 25 43 52 by PEGMA.
Base-MI − 28 42 50 The effect of thermal post-treatment on viscoelastic properties was
Base-Mix − 27 43 51
Base-3D − 30 41 46
investigated for Base-3D. As can be seen in Fig. 7, a slight increase in tan
delta peak, along with an increase in rubbery plateau (E ) is observed
′

during the first 2 h due to the network rearrangement caused by the

delta peak at lower temperature would correspond to the epoxy-rich transesterification. When 1-MI was used as catalyst, a post-treatment of
fraction (analysis of neat ER-GLU epoxy-acid polymer produced a Tg up to 4 h was required in order to obtain a stable structure, which
of ca. 50 ◦ C measured with DSC), while the second and broader peak suggests that Zn(AcAc)2 is a more effective catalyst for the

6
J. Casado et al. European Polymer Journal 173 (2022) 111256

Fig. 6. Storage moduli (E’) and tan delta curves of all formulations at curing stage 2.

Fig. 7. Effect of thermal post-treatment (at 180 ◦ C) on viscoelastic behavior of Base-3D. Inset: TGA curve at 180 ◦ C. No significant weight loss is observed in 24 h.

transesterification. In the light of these findings, a post-treatment altered as the reactions progress until reaching equilibrium. The
duration of 4 h is established for all samples studied. The material mobility of the network is slightly more restricted in the equilibrium
showed no significant thermal degradation at the post-treatment tem­ configuration, as was suggested by the shift of tan delta during the first 2
perature of 180 ◦ C (Fig. 7 inset). h of thermal post-treatment (see Fig. 7).

3.3. Bond exchange mechanisms 3.4. Stress relaxation behavior

The fully cured polyacrylate-epoxy dual network can undergo
transesterification reactions in a number of ways, some of which are
depicted in Scheme 2. An exhaustive scheme depicting all possible
transesterifications is not provided due to space restrictions. These ester-
hydroxyl exchanges can take place in any arbitrary combination
involving the β-hydroxy esters in i) the poly(acrylate), ii) the GMA-GLU
oligomer, and iii) the epoxy-acid (ER-GLU) polymer,
As can be seen from Scheme 2, overall network connectivity can be
Fig. 8 shows the effect of the thermal post-treatment on the relaxa­
tion of Base-3D fully cured material at 180 ◦ C. In post-treated samples,
relaxation is slowed down slightly due to a more densely cross-linked
structure after transesterification equilibrium is established (Fig. 7). As
observed before, 2 h are sufficient to reach an equilibrium structure and
repeatable relaxation kinetics are observed after 2 h of thermal post
treatment.
Fig. 9 shows the effect of temperature on the stress relaxation of post-

7
J. Casado et al. European Polymer Journal 173 (2022) 111256

Scheme 2. Possible ester-hydroxyl exchange reactions within the polyacrylate network (a) and in between polyacrylate and epoxy-acid networks (b).

Fig. 8. Stress relaxation of Base-3D post-treated for different times. Specimens were subjected to a strain of 1% in 3-point bending.

treated Base-3D samples. Relaxation profiles evidence that trans­
esterification hastens as temperature is increased: While at 160 ◦ C full
stress relaxation requires more than one day, at 200 ◦ C it is nearly
complete in 2 h. These results compare well with other
transesterification-based CANs reported in the literature [12,16].
Naturally, the choice of catalyst also affects stress relaxation
behavior, as seen in Fig. 10. When 1-MI was used as co-catalyst, relax­
ation rate slowed down by a factor of 2 to 3, and when 1-MI was used
singularly, transesterification rate decreased by a decade. The presence
of PEGMA did not seem to have any detrimental effect on the relaxation
kinetics.
The poor relaxtion behavior of the 1-MI catalyzed formulation was
surprising, since 1-MI is documented as an excellent transesterification
catalyst [25]. The effect observed herein was rationalized in terms of the
network structure and the role of 1-MI in the epoxy-acid reaction. (1)-MI
is known to have nucleophilic behavior, forming adducts and complexes
with the epoxide component which in turn propagate the curing
[32,33]. In our hybrid system, the epoxy-acid reaction is the second
curing reaction that takes place within the polyacrylate matrix formed
earlier. As a consequence, a significant fraction of 1-MI may be cova­
lently bound to the polyester network structure formed by the reaction
between ER (and GMA) and GLU. As suggested by the bimodal glass
transitions seen in Fig. 6 (except Base-3D), the hybrid network is not
entirely homogeneous. Consequently, one can argue that the part of 1-
MI segregated in the epoxy-acid network would be unavailable to
catalyze the transesterifications involving the β-hydroxy esters of the
polyacrylate network. On the other hand, Zn(AcAc)2 would be homo­
geneously distributed within the material and would have no preferred
interaction within a particular region of the network, therefore main­
taining its availability to catalyze both intra- and inter-network

8
J. Casado et al. European Polymer Journal 173 (2022) 111256

Fig. 9. Normalized stress relaxation curves of Base-3D at varying temperatures. The dashed line marks the E/E0 value of 1/e (i.e. at ~63% of complete relaxation)
used to calculate the vitrimeric transition temperature Tv.

Fig. 10. Stress relaxation of all formulations at 200 ◦ C, samples were post-treated for 4 h at 180 ◦ C prior to analysis.

transesterifications. In Base-Mix, 1-MI may have had an acid-base
interaction with Zn(AcAc)2, thereby neutralizing it partially, leading
to slower relaxation in comparison with the material containing only Zn
(AcAc)2 as catalyst.
To elaborate on relaxation kinetics, each formulation was tested at
four different temperatures to confirm Arrhenius-like behavior of the
characteristic relaxation times τ* (i.e. time required to relax ~63% of the
stress) which is expressed as τ(T) = τ0 ⋅exp(Eact /RT). As Base-MI material
exhibits longer relaxation times compared to other formulations, it was
tested at higher temperatures. The Arrhenius plots of the relaxation
times are shown in Fig. 11. The test temperatures are specified in the
figure caption. As can be deduced from the slope of the different lines,
the activation energies of the different systems are similar, regardless of
the network structure and the catalyst employed. The calculated values,
shown in Table 4, are comparable to other transesterification-based
CANs reported in literature [7,12,24].
The so-called liquid-to-solid transition temperature, Tv was deter­
mined for each formulation following the procedure outlined by Capelot
et al [24]. The Tv is defined as the temperature at which the viscosity
exceeds 1012 Pa.s. Below this temperature, transesterification rate is
assumed negligible. Applying the Maxwell equation given as η = G⋅τ*
together with the relation between shear modulus and tensile modulus
G = E’ /[2(1 +ν)] where ν is the Poisson’s ratio that attains a value of 0.5
in the rubbery state, we get G = E’/3. Taking E’ 3 MPa, τ* is calculated
as 106 s. Using this result, it is straightforward to determine Tv from the
plots in Fig. 11. It is evidenced from the graph that the Tv of Base-MI is

9
J. Casado et al. European Polymer Journal 173 (2022) 111256

Fig. 11. Arrhenius plots of the relaxation times. Base-Zn, Base-Mix and Base-3D were tested at 160, 180, 200 and 220 ◦ C whereas Base-MI was tested at 180, 200, 220
and 240 ◦ C.

Table 4
Calculated Eact and Tv of the different formulations, obtained from analysis of
samples post-treated at 180 ◦ C for 4 h.
Formulation Eact (kJ/mol) Tv (◦ C)

Base-Zn 94.8 133

Base-MI 116.8 177
Base-Mix 101.7 148
Base-3D 102.2 138

significantly higher than others. Indeed, below 180 ◦ C, Base-MI did not
relax within a practical timeframe. Actually, the Tv of Base-MI suggests
its relaxation rate would be negligible below 180 ◦ C, while relaxation
would occur for Base-Zn and Base-3D at temperatures as low as 140 ◦ C.
All these observations corroborate experimental findings presented
previously.

3.5. Printed, repaired and recycled materials

Fig. 12. Stress-strain plots of fully cured Base-3D samples. Inset: A zeolite
Complex objects can be DLP-printed easily and at rapid rates with structure and a model fountain printed with Base-3D.
Base-3D. Photographs of two such objects are given as inset in Fig. 12.
To test repairability, rectangular DMA samples were 3D printed using
finely chopped and later hot-pressed as explained in Materials and
Base-3D with the same dimensions as in earlier stress–strain tests. One of
Methods section. From the solid films obtained, new DMA samples were
these samples were printed with a hole at its center as can be seen in the
die-cut in the same dimensions, after which they were analyzed in DMA
legend of Fig. 12. This perforated sample exhibited a significantly lower
at the same conditions. As seen in Fig. 13, apart from a slight increase in
Young’s modulus (846 MPa) compared to the pristine sample (1323
storage modulus and a slight decrease of rubbery plateau, both of which
MPa). Later, it was repaired by simply filling the hole with either the
are practically insignificant, the recycled material is identical to its
liquid Base-3D, or a commercial acrylic resin with trade name Spot-LV
pristine counterpart. The inset in Fig. 13 illustrates the recycling
(Spot-A Materials, Barcelona, Spain) which has a similar final Tg but
procedure.
has no bond exchanging components (i.e. a CAN-free material). Both
repaired samples were fully processed using the established procedure
4. Conclusions
(i.e. Photocure -> thermal cure –> thermal post-treatment). Whereas
the stress-response of the sample repaired using Base-3D indicated full
We have described a new poly(hydroxy-ester) dynamic network
recovery, the sample repaired using Spot-LV exhibited only partial re­
obtained through a simple dual-cure procedure. The unique formulation
covery. The recycled sample exhibited a Young’s modulus of 1514 MPa,
combines a photocurable diacrylate already rich in β-hydroxy esters and
a value slightly higher than that of the original sample (1323 MPa)
hydroxyls, with an acid cured epoxy to yield a hybrid network capable of
probably due to volatilization of small fragments during the recycling
undergoing intra- and inter-network transesterifications. This is
process.
exploited in repairing and recycling these vitrimeric materials via facile
Recyclability was tested on DMA samples. Fully cured samples were

10
J. Casado et al.
Fig. 13. DMA storage modulus and tan delta curves of recycled samples in comparison with a pristine one. The recycling procedure is depicted as inset.
procedures.
Zinc acetylacetonate and 1-methylimidazole were tested as catalysts
both of the epoxy-acid reaction and transesterifications. At moderately
elevated temperatures, transesterification reactions are initiated across
the hybrid network at sufficiently high rates. Reaction equilibrium is
reached in a few hours, which brings about an apparent increase in
crosslinking density and Tg . When subjected to strain, these materials
are able to relax their stress within practical timeframes at temperatures
as low as 140 ◦ C. Their relaxation kinetics were comparable to other
vitrimers based on transesterification reactions.
The imidazole catalyst was more efficient in the catalysis of the
epoxy-acid reaction but showed a poor performance as trans­
esterification catalyst. The epoxy-acid reaction was not as fast when the
zinc catalyst was used, but it reached completion in a reasonable time.
This lower reactivity made the zinc catalyst more adequate in terms of
storage stability of the uncured formulations and intermediate stage
materials. Moreover, the transesterification was significantly faster with
the zinc catalyst. A combination of the two catalysts produced an in­
termediate effect, promoting fast epoxy-acid reaction while retaining an
acceptable rate of transesterification.
The formulation modified with PEGMA, thanks to its low viscosity,
could be used to print objects in a DLP type 3D printer. After thermal
curing and thermal post-treatment, the resulting fully-cured 3D-printed
objects were shown to be repairable and completely recyclable. By
combining processing flexibility of dual-cure formulations with the
versatility of DLP-based 3D printing methods, these novel formulations
show promise in advanced applications requiring tailor-made properties
and processing flexibility. Coupled with their repair and recycle capa­
bilities facilitated by a covalently adaptable network structure, they can
potentially replace conventional thermosets and reduce their carbon
footprint.
CRediT authorship contribution statement
J. Casado: Formal analysis, Investigation, Writing – original draft.
O. Konuray: Conceptualization, Investigation, Methodology, Supervi­
sion, Validation, Writing – original draft, Writing – review & editing. A.
Roig: . X. Fernandez-Fráncos: Conceptualization, Methodology, Su­
pervision, Validation, Writing – review & editing. X. Ramis: Concep­
tualization, Formal analysis, Funding acquisition, Investigation,
Methodology, Project administration, Resources, Supervision, Valida­
tion, Writing – review & editing.
11
European Polymer Journal 173 (2022) 111256
Declaration of Competing Interest
The authors declare that they have no known competing financial
interests or personal relationships that could have appeared to influence
the work reported in this paper.
Acknowledgements
The authors acknowledge the funding sources disclosed below. The
authors also thank Po.Int.Er S.R.L. for supplying the epoxy resin. X.
Fernández-Francos and O. Konuray acknowledge the Serra-Húnter
programme (Generalitat de Catalunya).
Funding
This work was funded by the Spanish Ministry of Science and Inno­
vation (MCNI/AEI) through R&D projects PID2020-115102RB-C21 and
PID2020-115102RB-C22, and also by Generalitat de Catalunya (2017-
SGR-77 and BASE3D).
Data availability
The raw data required to reproduce these findings are available upon
request from authors.
References
[1] C.J. Kloxin, T.F. Scott, B.J. Adzima, C.N. Bowman, Covalent adaptable networks
(CANs): A unique paradigm in cross-linked polymers, Macromolecules 43 (2010)
2643–2653, https://doi.org/10.1021/ma902596s.
[2] W. Denissen, J.M. Winne, F.E. Du Prez, Vitrimers: permanent organic networks
with glass-like fluidity, Chem. Sci. 7 (2016) 30–38, https://doi.org/10.1039/
C5SC02223A.
[3] C.J. Kloxin, C.N. Bowman, Covalent adaptable networks: Smart, reconfigurable and
responsive network systems, Chem. Soc. Rev. 42 (2013) 7161–7173, https://doi.
org/10.1039/c3cs60046g.
[4] W. Zou, J. Dong, Y. Luo, Q. Zhao, T. Xie, Dynamic Covalent Polymer Networks:
from Old Chemistry to Modern Day Innovations, Adv. Mater. 29 (2017) 1–18,
https://doi.org/10.1002/adma.201606100.
[5] B. Krishnakumar, R.V.S.P. Sanka, W.H. Binder, V. Parthasarthy, S. Rana, N. Karak,
Vitrimers: Associative dynamic covalent adaptive networks in thermoset polymers,
Chem. Eng. J. 385 (2020), 123820, https://doi.org/10.1016/j.cej.2019.123820.
[6] M. Podgórski, B.D. Fairbanks, B.E. Kirkpatrick, M. McBride, A. Martinez,
A. Dobson, N.J. Bongiardina, C.N. Bowman, Toward Stimuli-Responsive Dynamic
Thermosets through Continuous Development and Improvements in Covalent
J. Casado et al.
Adaptable Networks (CANs), Adv. Mater. 32 (2020) 1–26, https://doi.org/
10.1002/adma.201906876.
[7] D. Montarnal, M. Capelot, F. Tournilhac, L. Leibler, Silica-Like Malleable Materials
from Permanent Organic Networks, Science 334 (6058) (2011) 965–968.
[8] K. Yu, P. Taynton, W. Zhang, M.L. Dunn, H.J. Qi, Reprocessing and recycling of
thermosetting polymers based on bond exchange reactions, RSC Adv. 4 (2014)
10108–10117, https://doi.org/10.1039/c3ra47438k.
[9] X. Niu, F. Wang, X. Kui, R. Zhang, X. Wang, X. Li, T. Chen, P. Sun, A.C. Shi, Dual
Cross-linked Vinyl Vitrimer with Efficient Self-Catalysis Achieving Triple-Shape-
Memory Properties, Macromol. Rapid Commun. 40 (2019) 1–8, https://doi.org/
10.1002/marc.201900313.
[10] M. Capelot, D. Montarnal, F. Tournilhac, L. Leibler, Metal-catalyzed
transesterification for healing and assembling of thermosets, J. Am. Chem. Soc. 134
(2012) 7664–7667, https://doi.org/10.1021/ja302894k.
[11] L. Imbernon, S. Norvez, L. Leibler, Stress Relaxation and Self-Adhesion of Rubbers
with Exchangeable Links, Macromolecules 49 (2016) 2172–2178, https://doi.org/
10.1021/acs.macromol.5b02751.
[12] B. Zhang, K. Kowsari, A. Serjouei, M.L. Dunn, Q. Ge, Reprocessable thermosets for
sustainable three-dimensional printing, Nat. Commun. 9 (2018), https://doi.org/
10.1038/s41467-018-04292-8.
[13] C. Taplan, M. Guerre, F.E. Du Prez, Covalent Adaptable Networks Using β-Amino
Esters as Thermally Reversible Building Blocks, J. Am. Chem. Soc. 143 (2021)
9140–9150, https://doi.org/10.1021/jacs.1c03316.
[14] F. Cuminet, S. Caillol, É. Dantras, É. Leclerc, V. Ladmiral, Neighboring Group
Participation and Internal Catalysis Effects on Exchangeable Covalent Bonds:
Application to the Thriving Field of Vitrimer Chemistry, Macromolecules 54 (2021)
3927–3961, https://doi.org/10.1021/acs.macromol.0c02706.
[15] J. Han, T. Liu, C. Hao, S. Zhang, B. Guo, J. Zhang, A Catalyst-Free Epoxy Vitrimer
System Based on Multifunctional Hyperbranched Polymer, Macromolecules 51
(2018) 6789–6799, https://doi.org/10.1021/acs.macromol.8b01424.
[16] E. Rossegger, R. Höller, D. Reisinger, M. Fleisch, J. Strasser, V. Wieser, T. Griesser,
S. Schlögl, High resolution additive manufacturing with acrylate based vitrimers
using organic phosphates as transesterification catalyst, Polymer 221 (2021),
https://doi.org/10.1016/j.polymer.2021.123631.
[17] U. Shaukat, E. Rossegger, S. Schlögl, Thiol–acrylate based vitrimers: From their
structure–property relationship to the additive manufacturing of self-healable soft
active devices, Polymer 231 (2021) 29–34, https://doi.org/10.1016/j.
polymer.2021.124110.
[18] E. Rossegger, R. Höller, D. Reisinger, J. Strasser, M. Fleisch, T. Griesser, S. Schlögl,
Digital light processing 3D printing with thiol-Acrylate vitrimers, Polym. Chem. 12
(2021) 638–644, https://doi.org/10.1039/d0py01520b.
[19] Q. Shi, K. Yu, X. Kuang, X. Mu, C.K. Dunn, M.L. Dunn, T. Wang, H. Jerry Qi,
Recyclable 3D printing of vitrimer epoxy, Mater. Horizons. 4 (2017) 598–607,
https://doi.org/10.1039/c7mh00043j.
12
European Polymer Journal 173 (2022) 111256
[20] Z. Chen, M. Yang, M. Ji, X. Kuang, H.J. Qi, T. Wang, Recyclable thermosetting
polymers for digital light processing 3D printing, Mater. Des. 197 (2021) 189,
https://doi.org/10.1016/j.matdes.2020.109189.
[21] C. Lu, C. Wang, J. Yu, J. Wang, F. Chu, Two-Step 3 D-Printing Approach toward
Sustainable, Repairable, Fluorescent Shape-Memory Thermosets Derived from
Cellulose and Rosin, ChemSusChem 13 (2020) 893–902, https://doi.org/10.1002/
cssc.201902191.
[22] X. Fernández-francos, O. Konuray, X. Ramis, À. Serra, S. De, Enhancement of 3D-
printable materials by dual-curing procedures, Materials. 14 (2021) 1–23.
[23] O. Konuray, X. Fernández-Francos, X. Ramis, À. Serra, State of the Art in Dual-
Curing Acrylate Systems, Polymers. 10 (2018) 178, https://doi.org/10.3390/
polym10020178.
[24] M. Capelot, M.M. Unterlass, F. Tournilhac, L. Leibler, Catalytic control of the
vitrimer glass transition, ACS Macro Lett. 1 (2012) 789–792, https://doi.org/
10.1021/mz300239f.
[25] F.I. Altuna, C.E. Hoppe, R.J.J. Williams, Shape memory epoxy vitrimers based on
DGEBA crosslinked with dicarboxylic acids and their blends with citric acid, RSC
Adv. 6 (2016) 88647–88655, https://doi.org/10.1039/c6ra18010h.
[26] X. Fernández-Francos, S.G. Kazarian, X. Ramis, À. Serra, Simultaneous Monitoring
of Curing Shrinkage and Degree of Cure of Thermosets by Attenuated Total
Reflection Fourier Transform Infrared (ATR FT-IR) Spectroscopy, Appl. Spectrosc.
67 (2013) 1427–1436, https://doi.org/10.1366/13-07169.
[27] O. Konuray, A. Altet, J. Bonada, A. Tercjak, X. Fernández-francos, X. Ramis, Epoxy
Doped, Nano-scale Phase-separated Poly-Acrylates with Potential in 3D Printing,
Macromol. Mater. Eng. 2000558 (2021) 1–10, https://doi.org/10.1002/
mame.202000558.
[28] O. Konuray, A. Sola, J. Bonada, A. Tercjak, A. Fabregat-Sanjuan, X. Fernández-
Francos, X. Ramis, Cost-Effectively 3D-Printed Rigid and Versatile Interpenetrating
Polymer Networks, Materials. 14 (16) (2021) 4544.
[29] M. Morancho, X. Fernández-Francos, X. Ramis, Curing kinetics of dually-processed
acrylate-epoxy 3D printing resins, Thermochim. Acta. 701 (2021), https://doi.org/
10.1016/j.tca.2021.178963.
[30] S. Vyazovkin, C.A. Wight, KINETICS IN SOLIDS, Annu. Rev. Phys. Chem. 48 (1997)
125–149, https://doi.org/10.1146/annurev.physchem.48.1.125.
[31] F.I. Altuna, C.E. Hoppe, R.J.J. Williams, Epoxy vitrimers: The effect of
transesterification reactions on the network structure, Polymers. 10 (2018),
https://doi.org/10.3390/polym10010043.
[32] M. Pire, C. Lorthioir, E.K. Oikonomou, S. Norvez, I. Iliopoulos, B. Le Rossignol,
L. Leibler, Imidazole-accelerated crosslinking of epoxidized natural rubber by
dicarboxylic acids: A mechanistic investigation using NMR spectroscopy, Polym.
Chem. 3 (2012) 946–953, https://doi.org/10.1039/c2py00591c.
[33] T.N. Tran, C. Di Mauro, A. Graillot, A. Mija, Chemical Reactivity and the Influence
of Initiators on the Epoxidized Vegetable Oil/Dicarboxylic Acid System,
Macromolecules 53 (2020) 2526–2538, https://doi.org/10.1021/acs.
macromol.9b02700.